ferret 0.9.6 → 0.10.0

data/test/unit/utils/tc_priority_queue.rb

@@ -0,0 +1,106 @@
+require File.dirname(__FILE__) + "/../../test_helper"
+
+
+class PriorityQueueTest < Test::Unit::TestCase
+  include Ferret::Utils
+
+  PQ_STRESS_SIZE = 1000
+
+  def test_pq()
+    pq = PriorityQueue.new(4)
+    assert_equal(0, pq.size)
+    assert_equal(4, pq.capacity)
+    pq.insert("bword")
+    assert_equal(1, pq.size)
+    assert_equal("bword", pq.top)
+
+    pq.insert("cword")
+    assert_equal(2, pq.size)
+    assert_equal("bword", pq.top)
+
+    pq << "dword"
+    assert_equal(3, pq.size)
+    assert_equal("bword", pq.top)
+    
+    pq << "eword"
+    assert_equal(4, pq.size)
+    assert_equal("bword", pq.top)
+
+    pq << "aword"
+    assert_equal(4, pq.size)
+    assert_equal("bword", pq.top, "aword < all other elements so ignore")
+
+    pq << "fword"
+    assert_equal(4, pq.size)
+    assert_equal("cword", pq.top, "bword got pushed off the bottom of the queue")
+
+    assert_equal("cword", pq.pop())
+    assert_equal(3, pq.size)
+    assert_equal("dword", pq.pop())
+    assert_equal(2, pq.size)
+    assert_equal("eword", pq.pop())
+    assert_equal(1, pq.size)
+    assert_equal("fword", pq.pop())
+    assert_equal(0, pq.size)
+    assert_nil(pq.top)
+    assert_nil(pq.pop)
+  end
+
+  def test_pq_clear()
+    pq = PriorityQueue.new(3)
+    pq << "word1"
+    pq << "word2"
+    pq << "word3"
+    assert_equal(3, pq.size)
+    pq.clear()
+    assert_equal(0, pq.size)
+    assert_nil(pq.top)
+    assert_nil(pq.pop)
+  end
+
+  #define PQ_STRESS_SIZE 1000
+  def test_stress_pq
+    pq = PriorityQueue.new(PQ_STRESS_SIZE)
+    PQ_STRESS_SIZE.times do
+      pq.insert("<#{rand(PQ_STRESS_SIZE)}>")
+    end
+
+    prev = pq.pop()
+    (PQ_STRESS_SIZE - 1).times do
+      curr = pq.pop()
+      assert(prev <= curr, "#{prev} should be less than #{curr}")
+      prev = curr
+    end
+    pq.clear()
+  end
+
+  def test_pq_block
+    pq = PriorityQueue.new(21) {|a, b| a > b}
+    100.times do
+      pq.insert("<#{rand(50)}>")
+    end
+
+    prev = pq.pop()
+    20.times do
+      curr = pq.pop()
+      assert(prev >= curr, "#{prev} should be greater than #{curr}")
+      prev = curr
+    end
+    assert_equal 0, pq.size 
+  end
+
+  def test_pq_proc
+    pq = PriorityQueue.new({:less_than => lambda {|a, b| a.size > b.size}, :capacity => 21})
+    100.times do
+      pq.insert("x" * rand(50))
+    end
+
+    prev = pq.pop()
+    20.times do
+      curr = pq.pop()
+      assert(prev.size >= curr.size, "#{prev} should be greater than #{curr}")
+      prev = curr
+    end
+    assert_equal 0, pq.size 
+  end
+end

metadata

@@ -3,8 +3,8 @@ rubygems_version: 0.8.11
 specification_version: 1
 name: ferret
 version: !ruby/object:Gem::Version 
-  version: 0.9.6
-date: 2006-08-23 00:00:00 +09:00
+  version: 0.10.0
+date: 2006-08-22 00:00:00 +09:00
 summary: Ruby indexing library.
 require_paths: 
 - lib
@@ -34,341 +34,173 @@ files:
 - README
 - MIT-LICENSE
 - TUTORIAL
-- CHANGELOG
-- ext/ferret.c
-- ext/lang.c
-- ext/r_analysis.c
-- ext/r_doc.c
-- ext/r_index_io.c
-- ext/r_qparser.c
-- ext/r_search.c
-- ext/r_store.c
-- ext/r_term.c
-- ext/extconf.rb
-- ext/inc
+- ext/q_filtered_query.c
+- ext/priorityqueue.c
+- ext/hashset.c
+- ext/helper.c
+- ext/global.c
+- ext/bitvector.c
+- ext/hash.c
+- ext/fs_store.c
+- ext/posh.c
+- ext/except.c
+- ext/ram_store.c
+- ext/store.c
 - ext/analysis.c
-- ext/stopwords.c
 - ext/document.c
+- ext/search.c
+- ext/similarity.c
+- ext/stopwords.c
+- ext/array.c
+- ext/index.c
+- ext/mem_pool.c
 - ext/compound_io.c
-- ext/index_rw.c
-- ext/vector.c
-- ext/field.c
-- ext/term.c
-- ext/q_parser.c
-- ext/q_const_score.c
+- ext/q_prefix.c
+- ext/q_range.c
+- ext/q_phrase.c
+- ext/q_term.c
+- ext/sort.c
 - ext/q_boolean.c
+- ext/filter.c
+- ext/q_const_score.c
 - ext/q_match_all.c
-- ext/q_phrase.c
-- ext/q_filtered_query.c
-- ext/search.c
-- ext/dummy.exe
-- ext/q_fuzzy.c
 - ext/q_wildcard.c
-- ext/ind.c
-- ext/q_range.c
-- ext/q_multi_phrase.c
-- ext/q_prefix.c
+- ext/q_fuzzy.c
+- ext/q_multi_term.c
+- ext/q_parser.c
 - ext/q_span.c
-- ext/filter.c
-- ext/similarity.c
-- ext/q_term.c
-- ext/sort.c
-- ext/index_io.c
-- ext/fs_store.c
-- ext/ram_store.c
-- ext/store.c
-- ext/hashset.c
-- ext/array.c
-- ext/bitvector.c
-- ext/helper.c
-- ext/global.c
-- ext/hash.c
-- ext/except.c
-- ext/priorityqueue.c
-- ext/libstemmer.h
-- ext/libstemmer.c
-- ext/modules.h
-- ext/utilities.c
-- ext/api.h
-- ext/api.c
-- ext/header.h
-- ext/stem_ISO_8859_1_italian.c
-- ext/stem_UTF_8_portuguese.c
-- ext/stem_UTF_8_portuguese.h
-- ext/ferret.h
-- ext/stem_UTF_8_french.c
-- ext/stem_UTF_8_spanish.c
+- ext/ind.c
+- ext/term_vectors.c
+- ext/priorityqueue.h
+- ext/defines.h
+- ext/posh.h
+- ext/store.h
+- ext/hashset.h
+- ext/helper.h
+- ext/global.h
+- ext/bitvector.h
+- ext/document.h
+- ext/hash.h
+- ext/except.h
+- ext/threading.h
+- ext/similarity.h
+- ext/index.h
+- ext/win32.h
+- ext/analysis.h
+- ext/search.h
+- ext/mem_pool.h
+- ext/array.h
+- ext/ind.h
+- ext/lang.h
+- ext/stem_UTF_8_norwegian.c
+- ext/stem_UTF_8_danish.c
 - ext/stem_UTF_8_dutch.c
-- ext/stem_ISO_8859_1_italian.h
+- ext/stem_UTF_8_english.c
+- ext/stem_UTF_8_finnish.c
+- ext/stem_UTF_8_french.c
 - ext/stem_UTF_8_german.c
-- ext/stem_UTF_8_french.h
-- ext/stem_UTF_8_spanish.h
+- ext/stem_UTF_8_italian.c
+- ext/stem_UTF_8_portuguese.c
+- ext/stem_UTF_8_porter.c
+- ext/stem_KOI8_R_russian.c
+- ext/stem_UTF_8_russian.c
+- ext/stem_UTF_8_spanish.c
+- ext/stem_UTF_8_swedish.c
 - ext/stem_ISO_8859_1_english.c
+- ext/stem_ISO_8859_1_danish.c
+- ext/stem_ISO_8859_1_dutch.c
+- ext/stem_ISO_8859_1_finnish.c
+- ext/stem_ISO_8859_1_french.c
+- ext/stem_ISO_8859_1_german.c
+- ext/stem_ISO_8859_1_italian.c
 - ext/stem_ISO_8859_1_norwegian.c
-- ext/stem_UTF_8_porter.c
+- ext/stem_ISO_8859_1_porter.c
+- ext/stem_ISO_8859_1_portuguese.c
+- ext/stem_ISO_8859_1_spanish.c
+- ext/stem_ISO_8859_1_swedish.c
+- ext/stem_UTF_8_danish.h
 - ext/stem_UTF_8_dutch.h
+- ext/stem_UTF_8_english.h
+- ext/stem_UTF_8_finnish.h
+- ext/stem_UTF_8_french.h
 - ext/stem_UTF_8_german.h
-- ext/stem_ISO_8859_1_english.h
-- ext/stem_ISO_8859_1_norwegian.h
+- ext/stem_UTF_8_italian.h
+- ext/stem_UTF_8_norwegian.h
 - ext/stem_UTF_8_porter.h
-- ext/stem_ISO_8859_1_portuguese.c
-- ext/stem_UTF_8_russian.c
-- ext/stem_ISO_8859_1_spanish.c
-- ext/stem_ISO_8859_1_french.c
-- ext/stem_ISO_8859_1_portuguese.h
-- ext/stem_ISO_8859_1_dutch.c
+- ext/stem_UTF_8_portuguese.h
 - ext/stem_UTF_8_russian.h
-- ext/stem_KOI8_R_russian.c
-- ext/stem_ISO_8859_1_german.c
-- ext/stem_ISO_8859_1_spanish.h
-- ext/stem_ISO_8859_1_french.h
-- ext/stem_ISO_8859_1_porter.c
-- ext/tags
-- ext/stem_ISO_8859_1_dutch.h
-- ext/stem_UTF_8_finnish.c
+- ext/stem_UTF_8_spanish.h
+- ext/stem_UTF_8_swedish.h
 - ext/stem_KOI8_R_russian.h
+- ext/stem_ISO_8859_1_danish.h
+- ext/stem_ISO_8859_1_dutch.h
+- ext/stem_ISO_8859_1_english.h
+- ext/stem_ISO_8859_1_finnish.h
+- ext/stem_ISO_8859_1_french.h
 - ext/stem_ISO_8859_1_german.h
+- ext/stem_ISO_8859_1_italian.h
+- ext/stem_ISO_8859_1_norwegian.h
 - ext/stem_ISO_8859_1_porter.h
-- ext/stem_UTF_8_finnish.h
-- ext/stem_UTF_8_danish.c
-- ext/stem_UTF_8_swedish.c
-- ext/stem_UTF_8_danish.h
-- ext/stem_UTF_8_swedish.h
-- ext/stem_ISO_8859_1_finnish.c
-- ext/stem_UTF_8_italian.c
-- ext/stem_ISO_8859_1_finnish.h
-- ext/stem_UTF_8_italian.h
-- ext/stem_ISO_8859_1_swedish.c
-- ext/stem_ISO_8859_1_danish.c
-- ext/stem_UTF_8_english.c
-- ext/stem_UTF_8_norwegian.c
+- ext/stem_ISO_8859_1_portuguese.h
+- ext/stem_ISO_8859_1_spanish.h
 - ext/stem_ISO_8859_1_swedish.h
-- ext/stem_ISO_8859_1_danish.h
-- ext/stem_UTF_8_english.h
-- ext/stem_UTF_8_norwegian.h
-- ext/document.h
-- ext/store.h
-- ext/array.h
-- ext/priorityqueue.h
-- ext/hashset.h
-- ext/helper.h
-- ext/global.h
-- ext/bitvector.h
-- ext/analysis.h
-- ext/hash.h
-- ext/search.h
-- ext/similarity.h
-- ext/index.h
-- ext/except.h
-- ext/lang.h
-- ext/frtio.h
-- ext/w32_io.c
-- ext/nix_io.c
+- ext/utilities.c
+- ext/api.c
+- ext/api.h
+- ext/header.h
+- ext/libstemmer.c
+- ext/modules.h
+- ext/libstemmer.h
+- ext/extconf.rb
+- ext/inc
+- ext/ferret.h
+- ext/ferret.c
+- ext/tags
+- ext/r_analysis.c
+- ext/r_utils.c
+- ext/r_store.c
+- ext/r_index.c
+- ext/r_search.c
+- ext/r_qparser.c
+- ext/inc/threading.h
 - ext/inc/lang.h
-- ext/inc/except.h
 - lib/ferret.rb
-- lib/rferret.rb
-- lib/ferret/analysis.rb
-- lib/ferret/document.rb
+- lib/ferret_version.rb
+- lib/ferret/number_tools.rb
 - lib/ferret/index.rb
-- lib/ferret/search.rb
-- lib/ferret/query_parser.rb
-- lib/ferret/stemmers.rb
-- lib/ferret/store.rb
-- lib/ferret/utils.rb
-- lib/ferret/analysis/standard_tokenizer.rb
-- lib/ferret/analysis/word_list_loader.rb
-- lib/ferret/analysis/tokenizers.rb
-- lib/ferret/analysis/token.rb
-- lib/ferret/analysis/analyzers.rb
-- lib/ferret/analysis/token_filters.rb
-- lib/ferret/analysis/token_stream.rb
-- lib/ferret/document/document.rb
-- lib/ferret/document/field.rb
-- lib/ferret/index/fields_io.rb
-- lib/ferret/index/index.rb
-- lib/ferret/index/segment_merge_queue.rb
-- lib/ferret/index/segment_term_enum.rb
-- lib/ferret/index/term_info.rb
-- lib/ferret/index/field_infos.rb
-- lib/ferret/index/segment_reader.rb
-- lib/ferret/index/index_file_names.rb
-- lib/ferret/index/index_reader.rb
-- lib/ferret/index/term_infos_io.rb
-- lib/ferret/index/term_enum.rb
-- lib/ferret/index/term.rb
-- lib/ferret/index/segment_merge_info.rb
-- lib/ferret/index/segment_merger.rb
-- lib/ferret/index/segment_infos.rb
-- lib/ferret/index/term_doc_enum.rb
-- lib/ferret/index/term_buffer.rb
-- lib/ferret/index/multi_reader.rb
-- lib/ferret/index/term_vectors_io.rb
-- lib/ferret/index/index_writer.rb
-- lib/ferret/index/term_vector_offset_info.rb
-- lib/ferret/index/segment_term_vector.rb
-- lib/ferret/index/document_writer.rb
-- lib/ferret/index/compound_file_io.rb
-- lib/ferret/index/multiple_term_doc_pos_enum.rb
-- lib/ferret/search/query.rb
-- lib/ferret/search/non_matching_scorer.rb
-- lib/ferret/search/top_field_docs.rb
-- lib/ferret/search/weight.rb
-- lib/ferret/search/field_doc.rb
-- lib/ferret/search/hit_queue.rb
-- lib/ferret/search/sort.rb
-- lib/ferret/search/disjunction_sum_scorer.rb
-- lib/ferret/search/boolean_scorer.rb
-- lib/ferret/search/boolean_query.rb
-- lib/ferret/search/field_sorted_hit_queue.rb
-- lib/ferret/search/conjunction_scorer.rb
-- lib/ferret/search/top_docs.rb
-- lib/ferret/search/req_opt_sum_scorer.rb
-- lib/ferret/search/boolean_clause.rb
-- lib/ferret/search/score_doc_comparator.rb
-- lib/ferret/search/req_excl_scorer.rb
-- lib/ferret/search/explanation.rb
-- lib/ferret/search/index_searcher.rb
-- lib/ferret/search/term_scorer.rb
-- lib/ferret/search/filter.rb
-- lib/ferret/search/similarity.rb
-- lib/ferret/search/fuzzy_term_enum.rb
-- lib/ferret/search/sort_field.rb
-- lib/ferret/search/hit_collector.rb
-- lib/ferret/search/term_query.rb
-- lib/ferret/search/scorer.rb
-- lib/ferret/search/score_doc.rb
-- lib/ferret/search/prefix_query.rb
-- lib/ferret/search/fuzzy_query.rb
-- lib/ferret/search/spans.rb
-- lib/ferret/search/caching_wrapper_filter.rb
-- lib/ferret/search/filtered_term_enum.rb
-- lib/ferret/search/multi_term_query.rb
-- lib/ferret/search/query_filter.rb
-- lib/ferret/search/wildcard_query.rb
-- lib/ferret/search/wildcard_term_enum.rb
-- lib/ferret/search/phrase_query.rb
-- lib/ferret/search/exact_phrase_scorer.rb
-- lib/ferret/search/sloppy_phrase_scorer.rb
-- lib/ferret/search/phrase_positions.rb
-- lib/ferret/search/phrase_scorer.rb
-- lib/ferret/search/range_query.rb
-- lib/ferret/search/multi_phrase_query.rb
-- lib/ferret/search/filtered_query.rb
-- lib/ferret/search/sort_comparator.rb
-- lib/ferret/search/range_filter.rb
-- lib/ferret/search/field_cache.rb
-- lib/ferret/search/match_all_query.rb
-- lib/ferret/search/multi_searcher.rb
-- lib/ferret/search/spans/near_spans_enum.rb
-- lib/ferret/search/spans/span_first_query.rb
-- lib/ferret/search/spans/spans_enum.rb
-- lib/ferret/search/spans/span_not_query.rb
-- lib/ferret/search/spans/span_near_query.rb
-- lib/ferret/search/spans/span_or_query.rb
-- lib/ferret/search/spans/span_term_query.rb
-- lib/ferret/search/spans/span_query.rb
-- lib/ferret/search/spans/span_scorer.rb
-- lib/ferret/search/spans/span_weight.rb
-- lib/ferret/query_parser/query_parser.tab.rb
-- lib/ferret/stemmers/porter_stemmer.rb
-- lib/ferret/store/index_io.rb
-- lib/ferret/store/directory.rb
-- lib/ferret/store/buffered_index_io.rb
-- lib/ferret/store/fs_store.rb
-- lib/ferret/store/ram_store.rb
-- lib/ferret/utils/priority_queue.rb
-- lib/ferret/utils/bit_vector.rb
-- lib/ferret/utils/parameter.rb
-- lib/ferret/utils/weak_key_hash.rb
-- lib/ferret/utils/string_helper.rb
-- lib/ferret/utils/number_tools.rb
-- lib/ferret/utils/date_tools.rb
-- lib/ferret/utils/thread_local.rb
-- test/test_helper.rb
+- lib/ferret/document.rb
 - test/test_all.rb
-- test/unit/ts_document.rb
-- test/unit/ts_index.rb
-- test/unit/ts_store.rb
+- test/test_helper.rb
 - test/unit/ts_analysis.rb
+- test/unit/ts_store.rb
 - test/unit/ts_utils.rb
+- test/unit/tc_document.rb
+- test/unit/ts_index.rb
 - test/unit/ts_search.rb
 - test/unit/ts_query_parser.rb
-- test/unit/utils/rtc_thread.rb
-- test/unit/utils/rtc_priority_queue.rb
-- test/unit/utils/rtc_bit_vector.rb
-- test/unit/utils/rtc_parameter.rb
-- test/unit/utils/rtc_weak_key_hash.rb
-- test/unit/utils/rtc_string_helper.rb
-- test/unit/utils/rtc_date_tools.rb
-- test/unit/utils/rtc_number_tools.rb
-- test/unit/analysis/rtc_standard_tokenizer.rb
-- test/unit/analysis/rtc_letter_tokenizer.rb
-- test/unit/analysis/rtc_stop_filter.rb
-- test/unit/analysis/rtc_stop_analyzer.rb
-- test/unit/analysis/ctc_analyzer.rb
+- test/unit/utils/tc_bit_vector.rb
+- test/unit/utils/tc_priority_queue.rb
+- test/unit/utils/tc_number_tools.rb
 - test/unit/analysis/tc_analyzer.rb
-- test/unit/analysis/rtc_standard_analyzer.rb
-- test/unit/analysis/rtc_lower_case_filter.rb
-- test/unit/analysis/rtc_porter_stem_filter.rb
-- test/unit/analysis/rtc_lower_case_tokenizer.rb
-- test/unit/analysis/rtc_white_space_tokenizer.rb
-- test/unit/analysis/tc_token.rb
-- test/unit/analysis/ctc_tokenstream.rb
-- test/unit/analysis/rtc_white_space_analyzer.rb
-- test/unit/analysis/rtc_word_list_loader.rb
-- test/unit/analysis/rtc_per_field_analyzer_wrapper.rb
-- test/unit/index/rtc_term_buffer.rb
-- test/unit/index/rtc_fields_io.rb
-- test/unit/index/rtc_term_vectors_io.rb
-- test/unit/index/rtc_segment_infos.rb
-- test/unit/index/tc_index.rb
-- test/unit/index/rtc_compound_file_io.rb
-- test/unit/index/rtc_field_infos.rb
-- test/unit/index/tc_index_writer.rb
-- test/unit/index/th_doc.rb
-- test/unit/index/rtc_segment_term_enum.rb
-- test/unit/index/rtc_term_info.rb
-- test/unit/index/tc_term.rb
-- test/unit/index/tc_term_voi.rb
-- test/unit/index/tc_index_reader.rb
-- test/unit/index/rtc_term_infos_io.rb
-- test/unit/index/rtc_segment_term_docs.rb
-- test/unit/index/rtc_segment_term_vector.rb
-- test/unit/index/rtc_multiple_term_doc_pos_enum.rb
-- test/unit/store/rtm_store_lock.rb
+- test/unit/analysis/tc_token_stream.rb
 - test/unit/store/tm_store_lock.rb
 - test/unit/store/tc_fs_store.rb
 - test/unit/store/tc_ram_store.rb
-- test/unit/store/rtm_store.rb
 - test/unit/store/tm_store.rb
-- test/unit/store/rtc_ram_store.rb
-- test/unit/store/rtc_fs_store.rb
-- test/unit/document/tc_field.rb
-- test/unit/document/rtc_field.rb
-- test/unit/document/tc_document.rb
-- test/unit/query_parser/tc_query_parser.rb
-- test/unit/search/tc_fuzzy_query.rb
-- test/unit/search/tc_multi_searcher2.rb
+- test/unit/index/tc_index_reader.rb
+- test/unit/index/tc_index_writer.rb
+- test/unit/index/tc_index.rb
+- test/unit/index/th_doc.rb
 - test/unit/search/tc_index_searcher.rb
-- test/unit/search/tc_spans.rb
-- test/unit/search/tc_filter.rb
+- test/unit/search/tc_multi_searcher.rb
+- test/unit/search/tc_fuzzy_query.rb
 - test/unit/search/tc_sort.rb
+- test/unit/search/tc_filter.rb
+- test/unit/search/tc_spans.rb
 - test/unit/search/tc_sort_field.rb
-- test/unit/search/tc_multi_searcher.rb
-- test/unit/search/rtc_sort_field.rb
-- test/unit/search/rtc_similarity.rb
 - test/unit/search/tc_search_and_sort.rb
-- test/longrunning/tc_numbertools.rb
-- test/longrunning/tm_store.rb
-- test/benchmark/tb_rw_vint.rb
-- test/benchmark/tb_ram_store.rb
-- test/functional/thread_safety_index_test.rb
-- test/functional/thread_safety_test.rb
-- test/utils/number_to_spoken.rb
-- test/unit/analysis/data/wordfile
-- rake_utils/code_statistics.rb
-- ext/termdocs.c
+- test/unit/search/tm_searcher.rb
+- test/unit/query_parser/tc_query_parser.rb
 test_files: []
 
 rdoc_options: 
@@ -384,6 +216,13 @@ extra_rdoc_files:
 - TODO
 - TUTORIAL
 - MIT-LICENSE
+- ext/r_analysis.c
+- ext/r_utils.c
+- ext/r_store.c
+- ext/r_index.c
+- ext/r_search.c
+- ext/r_qparser.c
+- ext/ferret.c
 executables: []
 
 extensions: